International Journal of Scientific Engineering and Technology

Volume No.4 Issue No.3, pp : 146-148

(ISSN : 2277-1581)
01 March. 2015

A Comparative Study of Various Data Transformation Techniques in Data

Mining
Km. Swati, Dr. Sanjay Kumar
Department of Computer Science and Engineering, Jaipur National University, Jaipur, India
Abstract: This research paper presents a technique to select an
ideal transformation technique of original and transformed
features. The paper reviews about a comparative study of various
data transformation techniques used in data mining which
includes six types of transformation techniques - Wavelets,
Genetic Algorithm and Wrappers, Identity transform, Program
synthesis, Data refinement transformation, and Feature Selection
technique. The feature selection technique is considered best as
it utilizes Wavelets and Genetic Algorithm and Wrappers
methods that employ classification accuracy as its fitness
function. The selection of transformed features provides new
insight on the interactions and behaviors of the features. This
method is especially effective with temporal data and provides
knowledge about the dynamic nature of the process. The
comparative study from the feature selection technique
demonstrates an improvement in classification accuracy,
reduction in the number of rules, and decrease in computational
time.
Keywords: Data transformation ,wavelets, genetic algorithm
and wrappers, feature selection technique
Introduction:
Data transformation converts a set of data values from the data
format of a source data system into the data format of a
destination data system. It is the process of converting data from
one format (e.g. a database file or Excel sheet) to another.
Because data often resides in different locations and formats
across the enterprise, data transformation is necessary to ensure
data from one application or database is intelligible to other
applications and databases, a critical feature for applications
integration. In a typical scenario where information needs to be
shared, data is extracted from the source application or data
warehouse, transformed into another format, and then loaded into
the target location. The quality of knowledge extracted from a
data set can be enhanced by its transformation. Discretization and
filling missing data are the most common forms of data
transformation.
It is an integral part of data mining and knowledge discovery [1]
so often used in data mining. For example, data is often
normalized to improve the effectiveness of the learning
algorithms (clustering, neural networks), but the effects of data
transformation on classification accuracy and knowledge
discovery has been limited.
Transforming data allows for an increased understanding of the
data and discovery of new and interesting relationships between
features.

IJSET@2015

Data transformation removes noise from data and also

summarizes data. Data transformation operations, such as
normalization and aggregation are additional data preprocessing
procedures that would contribute towards the success of mining
process. The Data transformation techniques can be utilized with
temporal data to improve the quality of knowledge.
Applying these transformations increases classification accuracy
of the extracted knowledge, enhances understanding of the
behavior of the features which results in more generalizable
rules sets (i.e., reduction
of the number of rules), and decreases the computation time.
VARIOUS DATA TRANSFORMATION TECHNIQUES
A) Wavelets
Wavelet transformations were developed to express the
frequency domain and the time locality of an input function. The
fact that wavelets capture the temporal nature of the data is quite
essential.Wavelet transformations consist of a family of
functions [2].
The wavelet transform is a tool that divides up data, functions, or
operators into different frequency components and then studies
each component with a resolution matched to its scale [3].
Therefore, the wavelet transform is anticipated to provide
economical and informative mathematical representation of
many objects of interest [4]. Nowadays many computer software
packages contain fast and efficient algorithms to perform
wavelet transforms. Due to such easy accessibility wavelets have
quickly gained popularity both in theoretical research and in
applications. Above all, wavelets have been widely applied in
such computer science research areas as image processing,
computer vision, network management, and data mining.
B) Genetic Algorithm and Wrappers
A genetic algorithm is a search technique that is based on natural
systems. It is used as a classifier directly in computation and also
optimize the results. Most applications of genetic algorithm in
pattern recognition optimize some parameters in the
classification process [5].
Genetic algorithms has been applied to find an optimal set of
feature weights that improve classification accuracy. First, a
traditional feature extraction method such as Principal
Component Analysis (PCA) is applied, and then a classifier such
as k-NN (Nearest Neighbor Algorithm) is used to calculate the
fitness function for it [6], [7]. Combination of classifiers is
another area that Genetic Algorithms have been used to
optimize. It is also used in selecting the prototypes in the casebased classification.
Page 146

International Journal of Scientific Engineering and Technology

Volume No.4 Issue No.3, pp : 146-148
The second method of genetic algorithm to optimize the result
from the dataset is more effective to compute the accurate values
of observations of data by applying data mining techniques.
A wrapper is a method incorporating a search algorithm and a
learning classifier to define ideal feature subsets. The wrapper in
this model utilizes a genetic algorithm to produce possible
feature subsets and a decision tree to evaluate the quality of each
subset. The algorithm was selected because it is widely used as
well as it generates implicit knowledge in the form of rules.
Various parameters of the optimization can affect the
computational resources required to carry out the optimization.
Larger values or population size and number of iterations (see
runtime arguments) favor optimal solutions, but increase
computational time. This method performs aggressive feature
selection that optimizes cross-validation performance.
Additionally, it is capable of optimizing any performance
measure for any classifier type. Unfortunately, methods using
genetic algorithms tend to scale poorly with number of features
and training set size.
C) Identity Transform
The identity transform is a data transformation that copies the
source data into the destination data without change. The identity
transformation is considered an essential process in creating a
reusable transformation library.
By creating a library of variations of the base identity
transformation, a variety of data transformation filters can be
easily maintained.

(ISSN : 2277-1581)
01 March. 2015

variants of feature transformation are feature construction and

feature extraction. Both are sometimes called feature discovery.
Feature selection is a term commonly used in data mining to
describe the tools and techniques available for reducing inputs to
a manageable size for processing and analysis. Feature selection,
also known as variable selection, attribute selection or variable
subset selection, is the process of selecting a subset of relevant
features for use in model construction.
The main objective of feature selection is to choose a subset of
input variables by eliminating features, which are irrelevant or of
no predictive information. Feature selection techniques are a
subset of the more general field of feature extraction. Feature
extraction creates new features from functions of the original
features, whereas feature selection returns a subset of the
features.
A genetic wrapper feature selection algorithm is utilized to
identify the key features from each of the transformed data sets.
The selected features are then combined into a single data set.
The genetic wrapper selection method is applied to the combined
data set. Wrapper methods use a predictive model to score
feature subsets.
Each new subset is used to train a model, which is tested on a
hold-out set. Counting the number of mistakes made on that
hold-out set (the error rate of the model) gives the score for that
subset. As wrapper methods train a new model for each subset,
they are very computationally intensive, but usually provide the
best performing feature set for that particular type of model.

D) Program synthesis
Synthesis is a special form of automatic programming that is
most often paired with a technique for formal verification. The
goal is to construct automatically a program that provably
satisfies a given high-level specification. In contrast to other
automatic programming techniques, the specifications are
usually non-algorithmic statements of an appropriate logical
calculus.
Data refinement
Data refinement is used to convert an abstract data model (in
terms
of sets for
example)
into
implementable data
structures (such as arrays). Operation refinement converts
a specification of an operation on a system into an
implementable program (e.g., a procedure).
The post condition can be strengthened and/or the pre condition
weakened in this process. This reduces any non determinism in
the
specification,
typically
to
a
completely deterministic implementation.
Feature Selection technique
This technique identifies the best features as compared to others.
Several transformation schemes are applied to the original data
set. Feature selection is also useful as part of the data analysis
process, as it shows which features are important for prediction,
and how these features are related. Feature transformation is a
process through which a new set of features is created. The
Figure: Feature selection
IJSET@2015

Page 147

International Journal of Scientific Engineering and Technology

Volume No.4 Issue No.3, pp : 146-148

(ISSN : 2277-1581)
01 March. 2015

CONCLUSION

REFERENCES

Data mining offer methods and tools for discovery of new

knowledge for making decision support system. Classification
accuracy of decisions made with the extracted knowledge
depends on the properties of the training data set. In most data
mining applications raw data is used for rule and feature
extraction.
Feature Selection technique identifies the best features as
compared to others. The comparative study from the feature
selection technique demonstrates an improvement in
classification accuracy, reduction in the number of rules, and
decrease in computational time. The methods are useful for
dimension reduction when the transformed features have a
descriptive power that is more easily ordered than the original
features.
In this paper a feature selection transformation method is best
transformation technique. This transformation, when applied to a
training data set, enhances classification accuracy of the decision
rules generated from different-different set. The reason for
increased classification accuracy with feature selection might
due to the fact that the associations among features and decisions
are stronger than those built on feature values.

i.
Howlett, and L.C. Jain (Eds), Knowledge-Based
Intelligent Information and Engineering Systems, LNAI 3213, Vol. I,
Springer, Heidelberg, Germany, 2004, pp. 148-154.
ii.
Hubbard, B.B.: The World According to Wavelets:
The Story of a Mathematical Technique in the Making. 2nd edn. Peters,
A.K. (eds.) Natick, MA (1998)
iii.
I. Daubechies. Ten Lectures on Wavelets. Capital
City Press, Montpelier, Vermont, 1992.
iv.
F. Abramovich, T. Bailey, and T. Sapatinas.Wavelet
analysis and its statistical applications. JRSSD, (48):130, 2000.
v.
Pattern Recognition Letters, (1995).Vol. 16, pp.
801-808
vi.
Siedlecki, W., Sklansky J., A note on genetic
algorithms for large-scale feature selection, Pattern Recognition
Letters, Vol. 10, Page 335-347, (1989).
vii.
Pei, M., Punch, W.F., and Goodman, E.D. Feature
Extraction Using Genetic Algorithms.

IJSET@2015

Page 148